| Session Name: | AI Summit: 'Little Learning Machines' Postmortem: A Game About Training Neural Networks |
| Speaker(s): | Dante Camarena, Nick Counter |
| Company Name(s): | Transitional Forms Inc, Transitional Forms Inc |
| Track / Format: | AI Summit |
Did you know free users get access to 30% of content from the last 2 years?
Get your team full access to the most up to date GDC content
| Overview: | Recent developments in Reinforcement Learning (RL) have resulted in near-human ability in competitive games such as Go, StarCraft and Dota 2, at the cost of significant computing power. An often under-looked element of training RL is the satisfaction that researchers and enthusiasts gain while watching their agents evolve and learn.nnIn this talk, Co-Directors Nick Counter and Dante Camarena present the work done in constructing Little Learning Machines, a game designed to bring the experience of training your own AI to players of any skill level. They show how careful game design and tricky engineering can help create a game that makes training accessible to people without technical knowledge or expensive computers. This presentation is intended to inspire developers to push this idea further. |
Recommended
